[ntp:hackers] Two NTP servers with GPS date calculation errors
Steven Sommars
stevesommarsntp at gmail.com
Mon May 4 16:23:21 UTC 2015
TymServe 2100. I was told this model went into service in 1996 and that
support ended in 2010. See also
https://www.febo.com/pipermail/time-nuts/2015-May/091746.html
On Sun, May 3, 2015 at 2:47 PM, Hal Murray <hmurray at megapathdsl.net> wrote:
>
> stevesommarsntp at gmail.com said:
> > As discussed previously some NTP/GPS servers erroneously applied a leap
> > second on January 22, 2015. At 01:00 May 3, 2015 two of these servers
> > shifted the date by 1024 weeks and believe it is now Sept 17, 1995. The
> > change matches the GPS rollover size, but that event doesn't occur this
> > year.
>
> What brand/model of GPS receivers are doing this?
>
> Jan 22 is when they started announcing the next leap second. I have a
> GlobalSat MR350P that shifts by a second back-and-forth several times a
> week.
> I assume it's a firmware bug. (It uses a SiRF chip.)
>
> The 1024 week glitches don't have to line up with integral multiples of
> 1024
> weeks from the start time. If the firmware knows that the time is at least
> 500 weeks since the start, then the 1024 week band of valid times can go
> from
> 500 to 500+1024. You can get the "at least" time from a constant in the
> firmware (build time) or from the environment (file system or network).
>
>
> --
> These are my opinions. I hate spam.
>
>
>
>
More information about the hackers
mailing list